Q-1 University of Washington Computer Programming I Lecture 16: Sorting © 2000 UW CSE

Q-2 Overview Sorting defined Algorithms for sorting Selection Sort algorithm Efficiency of Selection Sort

Q-3 Sorting

Q-4 Sorting The problem: put things in order Usually smallest to largest: “ascending” Could also be largest to smallest: “descending” Lots of applications! ordering hits in web search engine preparing lists of output merging data from multiple sources to help solve other problems faster search (allows binary search) too many to mention!

Q-5 Sorting: More Formally Given an array b[0], b[1],... b[n-1], reorder entries so that b[0] <= b[1] <=... <= b[n-1] Shorthand for these slides: the notation array[i..k] means all of the elements array[i],array[i+1]...array[k] Using this notation, the entire array would be: b[0..n-1] P.S.: This is not C syntax !

Q-6 Sorting Algorithms Sorting has been intensively studied for decades Many different ways to do it! We’ll look at only one algorithm, called “Selection Sort” Other algorithms you might hear about in other courses include Bubble Sort, Insertion Sort, QuickSort, and MergeSort. And that’s only the beginning!

Q-7 Sorting Problem What we want: Data sorted in order

Q-8 Sorting Problem What we want: Data sorted in order sorted: b[0]<=b[1]<=…<=b[n-1] 0 n b

Q-9 Sorting Problem What we want: Data sorted in order sorted: b[0]<=b[1]<=…<=b[n-1] 0 n b Initial conditions

Q-10 Sorting Problem What we want: Data sorted in order sorted: b[0]<=b[1]<=…<=b[n-1] 0 n b unsorted 0 n b Initial conditions

Q-11 General situation Selection Sort

Q-12 General situation Selection Sort smallest elements, sorted 0 k n b remainder, unsorted

Q-13 General situation Selection Sort smallest elements, sorted 0 k n b remainder, unsorted Step: Find smallest element x in b[k..n-1] Swap smallest element with b[k], then increase k

Q-14 General situation Selection Sort smallest elements, sorted 0 k n b remainder, unsortedsmallest elements, sorted 0 k n b x Step: Find smallest element x in b[k..n-1] Swap smallest element with b[k], then increase k

Q-15 General situation Selection Sort smallest elements, sorted 0 k n b remainder, unsortedsmallest elements, sorted 0 k n b x Step: Find smallest element x in b[k..n-1] Swap smallest element with b[k], then increase k

Q-16 Subproblem: Find Smallest

Q-17 Subproblem: Find Smallest /* Find location of smallest element in b[k..n-1] */ /* Assumption: k < n */ /* Returns index of smallest, does not return the smallest value itself */

Q-18 Subproblem: Find Smallest /* Find location of smallest element in b[k..n-1] */ /* Assumption: k < n */ /* Returns index of smallest, does not return the smallest value itself */ int min_loc (int b[ ], int k, int n) { int j, pos; /* b[pos] is smallest element */ /* found so far */ pos = k; for ( j = k + 1; j < n; j = j + 1) if (b[j] < b[pos]) pos = j; return pos; }

Q-19 Code for Selection Sort /* Sort b[0..n-1] in non-decreasing order (rearrange elements in b so that b[0]<=b[1]<=…<=b[n-1] ) */ void sel_sort (int b[ ], int n) { int k, m; for (k = 0; k < n - 1; k = k + 1) { m = min_loc(b,k,n); swap(&a[k], &b[m]); }

Q b Example

Q b min_loc Example

Q b b min_loc Example

Q b b min_loc Example

Q b b b min_loc Example

Q b b b min_loc Example

Q b Example (cont.)

Q b min_loc Example (cont.)

Q b b min_loc Example (cont.)

Q b b min_loc Example (cont.)

Q b b b min_loc Example (cont.)

Q b b b min_loc Example (cont.)

Q-32 Example (concluded) b

Q-33 Example (concluded) b min_loc

Q-34 Example (concluded) b b min_loc

Q-35 Sorting Analysis How many steps are needed to sort n things? For each swap, we have to search the remaining array length is proportional to original array length n Need about n search/swap passes Total number of steps proportional to n 2 Conclusion: selection sort is pretty expensive (slow) for large n

Q-36 Can We Do Better Than n 2 ? Sure we can! Selection, insertion, bubble sorts are all proportional to n 2 Other sorts are proportional to n log n Mergesort Quicksort etc. log n is considerably smaller than n, especially as n gets larger As the size of our problem grows, the time to run a n 2 sort will grow much faster than a n log n one.

Q-37 Any better than n log n? In general, no. But in special cases, we can do better Example: Sort exams by score: drop each exam in one of 101 piles; work is proportional to n Curious fact: efficiency can be studied and predicted mathematically, without using a computer at all! This branch of mathematics is called complexity theory and has many interesting, unsolved problems.

Q-38 Comments about Efficiency Efficiency means doing things in a way that saves resources Usually measured by time or memory used Many small programming details have little or no measurable effect on efficiency The big differences comes with the right choice of algorithm and/or data structure

Q-39 Summary Sorting means placing things in order Selection sort is one of many algorithms At each step, finds the smallest remaining value Selection sort requires on the order of n 2 steps There are sorting algorithms which are greatly more efficient It’s the algorithm that makes the difference, not the coding details